Small RNA-Mediated Epigenetic Myostatin Silencing.

Myostatin (Mstn) is a secreted growth factor that negatively regulates muscle mass and is therefore a potential pharmacological target for the treatment of muscle wasting disorders such as Duchenne muscular dystrophy. Here we describe a novel Mstn blockade approach in which small interfering RNAs (siRNAs) complementary to a promoter-associated transcript induce transcriptional gene silencing (TGS) in two differentiated mouse muscle cell lines. Silencing is sensitive to treatment with the histone deacetylase inhibitor trichostatin A, and the silent state chromatin mark H3K9me2 is enriched at the Mstn promoter following siRNA transfection, suggesting epigenetic remodeling underlies the silencing effect. These observations suggest that long-term epigenetic silencing may be feasible for Mstn and that TGS is a promising novel therapeutic strategy for the treatment of muscle wasting disorders.

Introduction

Myostatin (Mstn) (formerly growth and differentiation factor 8) is an endogenous, secreted protein from the transforming growth factor-β superfamily that negatively regulates skeletal muscle growth and differentiation.[1] Mstn null mutations result in increased musculature in cattle,[2] mice[1,3] and humans.[4] Mstn inhibition has been shown to induce functional improvement in the mdx mouse model[5,6,7,8,9,10] of Duchenne muscular dystrophy, an X-linked, monogenic muscle wasting disorder caused by loss-of-function mutations in the gene encoding dystrophin. Impaired dystrophin function results in progressive muscle degeneration leading to death, typically in the third decade of life by cardiac or respiratory failure.[11] Previous studies have demonstrated that the combination of Mstn blockade with dystrophin restoration in mdx mice resulted in a greater functional improvement than either treatment alone.[12] Consequently, the Mstn signaling pathway is a promising pharmacological target for the treatment of muscle wasting conditions such as Duchenne muscular dystrophy. A number of strategies have been used to achieve Mstn blockade including Mstn neutralizing antibodies,[13] endogenous Mstn antagonists (Mstn propeptide,[5] follistatin,[7] and soluble Acvr2b (the Mstn receptor)),[14] destructive exon-skipping,[15] and RNA interference (RNAi).[16]

An alternative to the canonical RNAi pathway (also known as post-transcriptional gene silencing, PTGS) is transcriptional gene silencing (TGS).[17,18] TGS is a homology-dependent gene silencing pathway mediated by small interfering RNAs (siRNAs), viral/plasmid expressed short hairpin RNAs[19] or expressed antisense RNAs[20,21] with complementarity to target gene promoters. These small RNA effector molecules target low copy-number promoter-associated RNA transcripts[22] in order to recruit chromatin-remodeling factors[20] to the complementary promoter and, in some cases, induce promoter DNA methylation.[19,22] The induction of targeted epigenetic changes enables long-term therapeutic gene silencing. To this end Suzuki et al. showed that transient transfection of an siRNA targeting the human immunodeficiency virus-1 5′ long terminal repeat was able to suppress human immunodeficiency virus replication for 31 days in the HeLa-derived MAGIC-5 cells.[23] Building on this work, Yamagashi et al. were able to silence human immunodeficiency virus messenger RNA (mRNA) transcription and viral replication for up to 1 year using a retrovirus-expressed short hairpin RNAs in a T-cell line (Molt-4).[24] Similarly, Hawkins et al. showed that long-term (up to 31 days) transcriptional silencing of Ubiquitin C could be achieved by 3 days of tet-induced promoter-targeted short hairpin RNAs expression.[17] Collectively, these studies suggest that long-term, targeted gene suppression by TGS may be possible in a therapeutic context. The aim of this work is to demonstrate the feasibility of silencing of Mstn expression by TGS. We show that Mstn expression is silenced by a promoter-targeted siRNA and that the silencing involves epigenetic remodeling of the Mstn promoter. This study thus opens up a new therapeutic avenue in the treatment of muscle wasting disorders.

Results

Detection of Mstn promoter-associated RNA

Previous studies have shown that TGS in mammalian cells requires the presence of promoter RNA transcripts. The database of transcription start sites[25] and University of California, Santa Cruz genome browser[26] resources were used to identify the Mstn transcription start site. To characterize transcription at the Mstn promoter, directional reverse transcription polymerase chain reaction (RT-PCR) was performed using primers that amplify a 153 bp region upstream of the annotated transcription start site in a strand-specific manner (). Transcripts were detected in both sense and antisense orientations indicating the presence of promoter-associated RNA at the Mstn promoter. PCR amplicons were sequenced to confirm identity (data not shown). Reverse transcriptase minus (RT−) control PCR reactions failed to amplify ruling out genomic DNA contamination ().

Promoter-targeting siRNAs induce transcriptional silencing of Mstn

Four siRNAs targeting the Mstn sense promoter-associated RNA () were synthesized by in vitro transcription, transfected into differentiated C2C12 myotube cultures and Mstn expression assessed by reverse transcriptase-quantitative PCR (RT-qPCR). A mRNA-targeting siRNA that induces post-transcriptional gene silencing of Mstn was used as a positive control (PTGS control). One siRNA, siMstn-P2, was found to significantly reduce Mstn mRNA levels by 50% (). This level of silencing was observed in at least 20 independent transfections under similar conditions (i.e. 100 nM siRNA in differentiated myotube cultures) and found to be highly reproducible (mean knockdown = 48%, SD = 11.5%).

Reduction in Mstn expression was observed relative to two unrelated nonspecific control (NS ctrl) siRNAs (including an in vitro transcribed siRNA (siCCR5) which targets the human C–C chemokine receptor type 5) (Supplementary Figure S1a). In addition, transfection with two further control siRNAs; one with the siMstn-P2 sequence scrambled (siScrambled) and the other with the central four nucleotides of siMstn-P2 inverted (siMM), did not significantly reduce Mstn expression (Supplementary Figure S1b). Negligible batch-to-batch variation was observed between different siRNA preparations (Supplementary Figure S1c) and no significant cellular toxicity was observed between any of the siRNA treatments (Supplementary Figure S1d). Mstn silencing was further confirmed using a chemically synthesized siMstn-P2 (). In addition, promoter-targeted silencing was dose-dependent although significant Mstn knockdown was only observed at siRNA concentrations of 50 nM and 100 nM (). Conversely, maximal silencing by the PTGS control siRNA was observed at 10 nM (data not shown). Statistically significant knockdown of Mstn was also observed in H2K mdx cells (a murine myoblast cell line that carries a mutation in dystrophin exon 23)[27] indicating that the silencing effect is not restricted to the C2C12 line (). Taken together, these data suggest that Mstn is susceptible to siRNA-directed TGS.

Previous studies have suggested that delivery of siRNA to the nucleus is essential to induce TGS and that nuclear targeting peptides were required to facilitate this delivery.[18,20,28] However, in this study silencing was observed with both conventional, commercially available transfection reagents (INTERFERin and RNAiMax) and with a stearylated transportan-10 derived peptide (PepFect14) that had previously been shown to effectively deliver splice-switching oligonucleotides to the nucleus[29] ().

Given that TGS and PTGS occur via different mechanisms, we hypothesized that co-transfection of TGS and PTGS siRNAs would act in a combinatorial manner to improve maximal Mstn gene silencing. Transfection of a combination of 50 nM siMstn-P2 and 50 nM PTGS control siRNA was compared against 100 nM of each individual siRNA or a NS ctrl siRNA. The combination of the two siRNAs gave the greatest knockdown (87%) ().

Mstn silencing is independent of interferon induction

In contrast with chemically synthesized siRNA molecules, small RNAs generated by in vitro transcription from T7 promoters are tri-phosphorylated at the 5′ terminus and therefore have the potential to induce nonspecific knockdown by activating the interferon response.[30] In order to investigate this possibility, C2C12 myotube cultures were transfected with all relevant siRNAs. After 48 hours total RNA samples were reverse transcribed and levels of the interferon-induced genes 2′,5′ oligoadenylate synthase 1b and interleukin-6 were determined by RT-qPCR. Treatment with 15 μg/ml lipopolysaccharide was used as a positive control for interferon induction. Statistically significant induction of oligoadenylate synthase 1b and interleukin-6 was observed with the in vitro transcribed siRNAs (siMstn-P2 and siCCR5) but not with chemically synthesized siRNAs (). The stability of the reference gene transcript (β-Actin) was unaffected by transfection with in vitro transcribed siRNAs (). These results indicate that in vitro transcribed Silencer constructed siRNAs induce expression of interferon-stimulated genes whereas chemically synthesized siRNAs do not.

Epigenetic Mstn silencing

Acetylation of histone H3 lysine residues 9 and 27 is associated with transcriptionally active chromatin. Consequently, deacetylation of these residues is a necessary first step in the process of silent state chromatin formation. In order to determine whether siMstn-P2 induces epigenetic gene silencing, transfections were performed in the presence of the histone deacetylase inhibitor trichostatin A (TSA). siRNA transfected cultures were treated with a range of TSA concentrations (50 nM to 5 μM) and for each experimental condition Mstn expression was normalized to the NS ctrl. Silencing by siMstn-P2 was found to be sensitive to TSA concentrations above 500 nM whereas silencing by the PTGS control siRNA was largely unaffected (). Treatment with TSA was found to activate basal Mstn expression at low concentrations and was toxic at high concentrations, which is consistent with other reports[31] (). The observation that Mstn silencing by siMstn-P2 was abrogated by treatment with TSA at concentrations that activate Mstn expression in one case (500 nM) and are highly toxic in another (5 μM) is evidence that these factors are not confounding the results.

To investigate whether changes in chromatin structure are involved in Mstn silencing we performed chromatin immunoprecipitation analysis (ChIP) using antibodies against the silent state histone modifications; dimethyl-histone H3 lysine9 (H3K9me2) and trimethyl-histone H3 lysine27 (H3K27me3). Enrichment of H3K9me2 was detected at the Mstn promoter following treatment with siMstn-P2 although no change in H3K27me3 was detected (). Sensitivity of silencing to TSA treatment and changes in H3K9 methylation suggest that epigenetic remodelling at the Mstn promoter underlies the observed silencing effect.

Discussion

Here we have demonstrated that siRNAs complementary to the Mstn promoter induce silencing of Mstn mRNA expression. We have detected promoter-associated RNAs at the Mstn locus and the sense-orientation transcript is the predicted target for the siRNA (although we have not shown a direct association between these two RNA species). Silencing was abrogated by treatment with the histone deacetylase inhibitor, TSA. Conversely, the PTGS control siRNA, which silences Mstn by acting upon the mature mRNA, was unaffected by TSA treatment. These results suggest that silencing with a promoter-targeting siRNA occurs by a different gene silencing pathway from conventional RNA interference. The results are consistent with other studies of TGS although we have not shown direct evidence of silencing at the level of transcription.

We also observed that the combination of TGS and PTGS siRNAs resulted in higher levels of silencing than either strategy alone suggesting that the two silencing pathways can operate in a combinatorial manner. We and others have shown that, following transient transfection of TGS effectors, target genes are typically silenced by ~50% which is considerably less than with conventional RNAi. However, the level of silencing is expected to increase over time as the chromatin at the target locus becomes progressively more compacted and the promoter DNA methylated. These silencing kinetics were observed in studies that looked at long-term knockdown by TGS.[17,24,32] However, in the case of Mstn, even relatively low levels of silencing (27% protein reduction) have been shown to result in significant functional changes in muscle mass (10% increase) suggesting that high levels of silencing are not required for a therapeutically relevant effect.[16] The demonstration of protein level silencing will be required to advance TGS as a therapy although, in this study, we were unable to consistently detect Mstn protein in cultured cells by western blot.

Several of the early studies to report TGS in mammalian cells used siRNAs produced by in vitro transcription.[18,20,28] siRNAs produced in this manner have the potential to elicit off-target gene silencing via induction of the interferon response.[30] These nonspecific events are either due to a gene being directly regulated by interferons, or as a result of global mRNA down-regulation, following activation of RNase L. We investigated this phenomenon by measuring expression of two interferon-induced genes (2′,5′ oligoadenylate synthase 1b and interleukin-6) following siRNA transfections in order to ensure specific Mstn silencing by siMstn-P2. We have shown that in vitro transcribed siRNAs induce highly statistically significant increases in expression of interferon-stimulated genes. However, the findings that; (i) silencing is observed with a synthetic siMstn-P2 in the absence of induction of the interferon-stimulated genes, (ii) silencing occurs relative to an in vitro transcribed control siRNA (siCCR5) and (iii) the stability of the housekeeping transcript (β-Actin) is unaffected by transfection with in vitro transcribed siRNAs, suggests that Mstn inhibition is not the result of off-target silencing because of interferon induction and activation of RNase L, but is rather a target-specific silencing effect.

Promoter-specific silencing was reversed in the presence of the histone deacetylase inhibitor, TSA, consistent with previous studies that have shown that histone deacetylase 1 is required for TGS.[17,32] Similarly, we have detected a modest enrichment of the silent state chromatin modification, H3K9me2, following transfection with a Mstn promoter-targeting siRNA as reported previously.[17,20,24,28,32] These results are indicative of an epigenetic silencing process. Interestingly, enrichment of H3K27me3 was not detected. Hawkins et al. have shown by RNAi depletion experiments that Enhancer of zeste homologue 2, a histone methyltransferase that methylates histone H3 lysine27, is not essential for TGS. This suggests that epigenetic silencing can occur in the absence of H3K27 methylation and that there is a degree of redundancy between silent state histone modifications.[17]

The detection of a promoter transcript involved in epigenetic regulation of the Mstn locus suggests that this may constitute part of an endogenous mechanism of gene regulation. Several studies have identified endogenous small RNAs (microRNAs) with complementarity to target gene promoters that induce TGS.[33,34,35] Long antisense RNAs have also been linked to epigenetic control of gene expression[36,37] and have been found to bind chromatin-remodeling factors.[38] Recently, a long noncoding RNA has also been implicated in myogenesis by acting as a competitive inhibitor for microRNAs involved in muscle differentiation.[39] However, the role of long noncoding RNAs, microRNAs, and pRNAs in epigenetic silencing is still not well understood.

This study adds to the growing literature that suggests small RNAs can direct TGS in mammalian cells and confirms that TGS occurs in mouse cells consistent with previous reports.[40] Several studies have demonstrated TGS in the absence of epigenetic changes, so-called antigene TGS.[41,42] These studies generally use oligonucleotides targeting transcriptional start sites and are speculated to involve direct interaction between the oligonucleotide and chromosomal DNA. Our data are more consistent with studies which have found a role for epigenetic changes involved in silencing and therefore lend credence to the idea that there are at least two distinct TGS mechanisms (epigenetic-TGS and antigene-TGS).

The use of TGS in animal models has so far been limited to three studies. Turunen et al. demonstrated that TGS of vascular endothelial growth factor is possible in mouse muscle following lentiviral transduction.[40] Similarly, local injection of promoter-targeting siRNAs against the human papillomavirus E6/E7 and thioredoxin-interacting protein promoters results in TGS in mouse xenograft tumors[43] and rat retina,[44] respectively. Although these studies are highly promising, the general application of TGS in vivo has not yet been fully explored and may present its own unique obstacles. For instance, the relative abundance of promoter transcripts is likely to differ between cell culture models and live animals and may affect the efficacy of a TGS approach. Despite the paucity of in vivo studies the therapeutic application of TGS is a promising alternative to RNAi, which is dependent on the presence of RNAi effectors that are gradually degraded within the cell or dispersed between daughter cells following cell division (effectively halving their concentration). Repeat administration is therefore required to elicit long-term silencing by siRNAs. Conversely, TGS effectors induce long-term silencing that is inherited following cell division and maintenance of silencing is not dependent on the presence of effector molecules.[17] A single treatment (or short course of treatments) may be sufficient to induce long-term gene silencing, hence repeat administration of TGS effectors is not required.[17,23] Consequently, TGS-based therapies may require a small number of high doses to elicit an effect. The overall dose will thus be lower with TGS-based therapies as opposed to conventional RNAi, meaning lower toxicity of treatments and greatly reduced material costs. In addition, it has been shown that saturation of endogenous RNA processing pathways can lead to acute toxicity in vivo.[45] As TGS does not require repeat administration this risk is greatly reduced. Epigenetic silencing of Mstn by TGS is therefore a promising novel therapeutic strategy in the treatment of muscle wasting disorders.

Materials and Methods

Directional RT-PCR. Total RNA was extracted from C2C12 myotubes using TRIzol reagent (Invitrogen, Paisely, UK) as according to the manufacturer's protocol and DNase treated using the TURBO DNase-free kit (Ambion/Applied Biosystems, Warrington, UK). 1 μg of total RNA was then reverse transcribed using SuperScript III RT (Invitrogen). In order to differentiate between sense and antisense transcripts complementary DNA (cDNA) synthesis was primed with either MstnPro-Rev (5′-AGCTTGCCCTCGACTGTAAC) or MstnPro-Fwd (5′-TCCAAGTGGCTTTTTATATTCCA), respectively. Following reverse transcription, PCR was performed with both primers and the amplification products analyzed by agarose gel electrophoresis. Mock reverse transcription controls (RT−) were performed to demonstrate that product amplification was not because of genomic DNA contamination. PCR products were gel extracted and sequenced to confirm identity.

siRNA. siRNAs were produced by in vitro transcription using the Silencer siRNA construction kit (Ambion) as according to manufacturer's instructions or by chemical synthesis (Eurogentec, Seraing, Belgium). Chemically synthesized siRNAs had 3′ dTdT overhangs. siRNAs designed to target the Mstn sense promoter-associated RNA all conform to the AA(N19) pattern. All siRNA sequences used in this study are listed in Supplementary Table S1. Unless otherwise stated NS ctrl refers to “control siRNA duplex negative control” (Eurogentec, catalogue #SR-CL000-005). In addition, an siRNA which targets human C–C chemokine receptor type 5 (CCR5) mRNA was used as an additional NS ctrl produced by in vitro transcription. The sequence of the PTGS control siRNA was a kind gift from Luis Garcia.

Cell culture. C2C12 mouse myoblasts were maintained in growth media; Dulbecco's modified Eagle's media supplemented with 15% fetal bovine serum and antibiotic/antimycotic (all Invitrogen) and 5% CO2. For differentiation, 24 well cell culture plates were coated with 0.02% gelatin (Sigma-Aldrich, Dorset, UK) and UV sterilized. C2C12 myoblasts were seeded (5 × 104 cells/ml) in growth medium and grown overnight. When confluent, the cells were switched to differentiation media; Dulbecco's modified Eagle's media supplemented with 5% horse serum (Invitrogen) and antibiotic/antimycotic. Cells were cultured in differentiation media for seven days in order to induce differentiation to myotubes.

H2K mdx tsA58 mouse myoblasts[27] were cultured in Dulbecco's modified Eagle's media supplemented with 20% fetal bovine serum), 2% chick embryo extract (PAA Laboratories Ltd, Yeovil, UK) and 20 U/ml γ-interferon (Invitrogen) at 33 °C and 10% CO2 When confluent, the H2K mdx cells were switched to differentiation medium and switched to 37 °C and 5% CO2 for 7 days.

Cells were transfected with RNAiMax (Invitrogen) and INTERFERin (PolyPlus Transfection, Strasbourg, France) were performed as according to manufacturer's instructions. Peptide transfections with PepFect 14 (CePep, Stockholm, Sweden) were performed as follows. Separate peptide and siRNA solutions were prepared in sterile water. The mixes were combined in a ratio of 30:1 (peptide:siRNA), mixed gently and incubated for 30 minutes at 37 °C to allow transfection complexes to form. After incubation, the transfection mixtures were diluted in optiMEM (Invitrogen) to a total volume of 0.45 ml per well (24 well plate). The media was removed from the C2C12 cultures and the cells washed with phosphate buffered saline (Invitrogen). Transfection complexes in optiMEM were then added to each well as appropriate and the cultures returned to the incubator. After 1 hour the cultures were supplemented with 50 μl fetal bovine serum and after 24 hours the transfection mixture was removed and replaced with fresh media. For all experiments mock transfections received transfection reagent/peptide only. Trichostatin A and lipopolysaccharide (both Sigma-Aldrich) were added to cell culture medium as appropriate. Cell viability was determined using the MTS assay (Promega, Southampton, UK) as according to manufacturer's instructions.

RT-qPCR. RT-qPCR was performed as follows. The planning, implementation and reporting of RT-qPCR experiments were designed to comply with the Minimum Information for publication of Quantitative real-time PCR Experiments guidelines[46,47] as much as was possible or appropriate. To generate cDNA 2 μg of high quality total RNA was reverse transcribed using the High-Capacity cDNA Synthesis kit (Applied Biosystems) as according to manufacturer's instructions. 20 μl RT reactions were primed using random primers. Samples were incubated at 25 °C for 10 minutes to allow primer extension to occur and then cDNA was reverse transcribed for 2 hours at 37 °C. The reaction was terminated at 80 °C for 15 minutes. Typically 200 ng cDNA was added to each reaction such that a maximum of 2 µl of RT reaction was added per well. RT-qPCR was performed on a StepOne Plus Real-Time PCR System (Applied Biosystems) using TaqMan Gene Expression Mastermix (Applied Biosystems). Relative quantities of target mRNA were determined using the relative standard curve method. Data were analyzed using the StepOne Software v2.1 (Applied Biosystems). Standard curves were produced from serial dilutions of cDNA reverse transcribed from untreated experimental samples. Universal cycling conditions were used (95°C for 10 minutes (hotstart) and then 40 cycles of 95°C for 15 seconds, 60°C for 1 minute). Typically, gene of interest expression was normalized to β-Actin expression. TaqMan assays were purchased from Applied Biosystems and are listed in Supplementary Table S2. Validation of RT-qPCR assays is described in Supplementary Materials and Methods and Supplementary Figures S2 and S3.

ChIP. ChIP was performed as described previously with several modifications.[17] A full protocol is described in the Supplementary Materials and methods. Briefly, C2C12 myotubes were cultured in 10 cm dishes and siRNAs transfected with RNAiMax. After 48 hours cells were harvested for ChIP assays. Chromatin samples were formaldehyde cross-linked, sonicated to shear the chromatin DNA and histone complexes and immunoprecipiated with antibodies against H3K9me2 (#07-441, Upstate/Millipore, Billerica, MA) and H3K27me3 (#97565, Cell Signalling Technologies, Danvers, MA). DNA was recovered by phenol chloroform extraction and resuspended in 30 μl of nuclease free water. Quantitative PCR was performed on a Mastercycler Realplex real-time thermal cycler (Eppendorf, Hauppauge, NY) using KAPA SYBR Fast SYBR mastermix (KAPA Biosystems, Cambridge, MA) and the primers; MstnChIP-Fwd (5′-AGATTCATTGTGGAGCAGGAG) and MstnChIP-Rev (5′-ATATTAGTGCATGTACCGTCCG). The relative standard curve method was used to compare samples and a five fold dilution of input chromatin DNA used to prepare the standard curve (Supplementary Figure S4a). Dissociation curve analysis confirmed that only a single amplification product was generated by the PCR reaction (Supplementary Figure S4b). Sample values were calculated as background subtracted, fraction of input for each experimental group and then normalized to the NS ctrl siRNA group.

Statistical analysis. All statistical analysis was performed in SPSS v11.5 (IBM, Armonk, NY, USA). Data were assessed for normalcy using the Shapiro–Wilk test and equality of variance by the Levene's test. Significance within a data set was determined by one-way ANOVA (analysis of variance) and significant differences between treatment groups were determined by post hoc analysis. The Bonferroni correction or the Games–Howell test were used in the instances of equality and inequality of variance respectively. P values less than 0.05 were considered significant. Unless otherwise indicated all statistical comparisons are made against NS ctrl siRNA treated cultures (grey bars).

Figure S1. Validation of specific myostatin silencing. Figure S2. Standard curves for (a) myostatin, and (b) β-Actin (ACTB) reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR) assays. Figure S3. Reverse transcriptase minus controls for myostatin (Mstn) and β-Actin (ACTB) reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR) assays. Figure S4. Validation of quantitative polymerase chain reaction (qPCR) for chromatin immunoprecipitation analysis. Table S1. Sequences of small interfering RNAs used in this study. Table S2. Reverse transcriptase-quantitative polymerase chain reaction TaqMan Primer/Probe assays used in this study. Materials and Methods.